• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

PS5 vs Xbox Series X ‘Secret Sauce’ – SSD Speed And Velocity Architecture

DForce

NaughtyDog Defense Force
This is a lie. That is the point a good portion of posters have been trying to argue, generally by misinterpreting how SSDs and NAND actually works. It hasn't been so much some people downplaying SSDS, so much as anyone attempting to be a realist with regard to the SSDs is automatically viewed by some others as downplaying the SSDs.

It's not a lie.

People were posting that Spider-Man GDC video several times just to show people how SSDs will improve next gen graphics.

I literally replied to a poster who made the claim and explained to him how he was misinterpreting what people were saying.

Let's face it; after the Road to PS5 presentation (and I don't like to do this but a spade's a spade) a big flock of Sony fans on the forums who were obsessed to high hell over Teraflops (even when multiple posters, myself included, were trying to tell them Teraflops didn't mean everything), silently conceded that front. They began downplaying the Teraflop difference between the two systems (not in terms of percentages per-se, but in what the extra TF advantage on XSX can actually be utilized for) while hyping up the SSD and audio since Sony focused on those in particular with their presentation and gave specs that, on paper, seemed more impressive than MS's in those areas. It allowed those people to shift the narrative to the SSD, audio, I/O etc. while similarly creating a fake narrative of XSX "brute forcing" a solution and PS5 was the system pushing elegant optimizations, conveniently cutting out any focus on MS's deliberate optimizations and customizations with the XSX to further push this fake narrative.

The thing is, XboxGAF have been accusing Digital Foundry, NX Gamer etc. of downplaying TF.

If the PS5 was 12.2TF and the XSX was 12.1, people would still be talking about the SSD and how much it would improve games.

I wouldn't call it a narrative shift since people are basically learning from what DF and others are teaching them.


All the while, many of these same people continue to over-inflate the SSDs in terms of being a game-changing technology or paradigm shift, and completely downplay the GPGPU performance edge XSX has over PS5 (or pretend it doesn't exist at all and that the extra GPU throughput of XSX will only go to resolution, ignoring the ML texture upscaling in the GPU built to free up heavy expenditure of GPU resources on processing raw pixels through to the display, freeing up processing power for other tasks). When you try telling them that NAND has limits inherent to the technology that will prevent granularity of asset data for streaming in a way similar to volatile RAM, somehow that gets lumped into "downplaying the PS5 SSD", even though this affects both systems. Same if you bring up questions regarding the random write speeds, latency figures, page and block sizes, etc.

We are seemingly allowed to speculate on Sony using tech from other departments of their company as R&D foundations for potential PS5 features, but doing the same with MS regarding XSX is considered being a fanboy, wishful thinking, or foolish...even though they have already admitted to members of the Surface team working on the Xbox team. All the same, some strongly pro-Sony people who obsess over customizations on PS5 do not provide any leeway to entertain similar customizations conceptually being present on XSX, but expect strong pro-Microsoft people to bend the knee and do so when it comes to XSX features potentially being present on PS5. And all of this leads to disingenuous, lopsided, biased takes and discussions in next-gen speculation because there are a group of people who put out a false image of wanting the best for both systems but secretly only want their preferred platform to "win", even if that means generating fake narratives.

Yes there are some Xbox people who do this but from what I've noticed it is not to the same degree as the Sony fans engaging in similar tactics on the forum (as just one example). Now that might be going a bit beyond your point here but it needs to be stressed that claiming "When people tried to explain how SSDs will work, people just started saying, "It's not going to close the power gap in consoles" when that wasn't the point people on here were trying to make." is in fact demostrably false when keeping in mind the long-term discussion that's been prevalent for months by now on these systems.

There absolutely have been people trying to imply this very thing, maybe not directly and often layered in subtext, but it's been an idea fostered for a good bit by now. Again, predicated on things like "secret sauce" like the way this article words it, which is irresponsible considering so many important aspects of the systems are not even divulged yet. But that's all I want to say on that and thought what you mentioned was a good time to segway into it. Hopefully people get what I'm saying here.


CPU performance is just under 3%
GPU performance is around 15%
Memory is going to depend how it's used. Will PS5's setup be different than XsX, whic will allow more ran to be use dfor games? Still remains to be seen.


So that leaves the SSD, which is about 126% difference

From what I can tell, XboxGAF is treating the GPU and CPU specs are as something HUGE, but somehow think a 126% increase of raw power will not do much for next gen games?




Now that might be going a bit beyond your point here but it needs to be stressed that claiming "When people tried to explain how SSDs will work, people just started saying, "It's not going to close the power gap in consoles" when that wasn't the point people on here were trying to make." is in fact demostrably false when keeping in mind the long-term discussion that's been prevalent for months by now on these systems.

Again, this is BS.


People didn't know how fast the SSD was in comparison to the XSX because there was no conformation. When more people started to speak out (NX Gamer, Moore's Law is Dead etc.) that's when people starts to fully realize what the SSDs were capable of.


I've seen people make those claims in that thread and it's clearly not a lie if you were paying attention.
 
What will be the actual real world performance advantage that PS5 ssd will offer compared to series X. Is it just 1 second loading times faster? NVMEs are 7 times faster than regular ssd but loading times are just one second faster.

With the series X its very easy to explain. The graphics fidelity will be the best with series X having the best resolution, graphical effects and ray tracing. Can someone explain this in real world performance what PS5 ssd will offer.

PS5 won't have loading screens. This video doesn't apply.
 

killatopak

Member
What will be the actual real world performance advantage that PS5 ssd will offer compared to series X. Is it just 1 second loading times faster? NVMEs are 7 times faster than regular ssd but loading times are just one second faster.

With the series X its very easy to explain. The graphics fidelity will be the best with series X having the best resolution, graphical effects and ray tracing. Can someone explain this in real world performance what PS5 ssd will offer.

The stuff the new consoles can do isn’t just centered around load times this time. That goes for both consoles. We can’t exactly know the difference without experiencing next gen exclusive multiplats.

Really the only game remotely close to showing that bandwidth and fast seek speeds matter is Star Citizen.
 

DForce

NaughtyDog Defense Force
because after the hdd has loaded the game into ram, it's advantage ends. All this SSD, SSD, SSD shit, have you been living under a rock?
That's false.

ps5-slides-06-640x360.png


"Contains data for next 1 second of gameplay"

That means the data is not just sitting there, it's being streamed in and quickly dumped for more data.

The objective is to feed the memory data as fast as possible in a fraction of a second. If you scan stream more data within that fraction of a second, then you can clearly stream is higher quality assets within that time frame due to storage speeds.
 

Three

Member
Honestly, Microsoft has been on top of compression for some time now. It's also mentioned that Microsoft has dedicated hardware for Compression / Decompression. With the datasets below factoring in hardware bassed compression for the cloud in a small PCI form factor, I'm sure including this technology in the XSX will help massive datasets move in a short period of time. I'm speculating that maybe Microsoft doesn't need a ton of bandwidth to accomplish what Sony is. This could also be the reason why Phil mentions that the developers can get 100GB instantly. Microsoft was able to accomplish this without CPU latency at PCIe 3.0 speeds.



EDIT: I'll certainly be posting this in other SSD threads as no one else seems to have ever looked at, or knew of this technology.
That compression is not limited by read/write speeds and is for the cloud. MS have already stated their bandwidth WITH compression as 4.8Gb/s.

The 100GB that MS mention is space reserved for the application, not bandwidth.
 
Last edited:

DForce

NaughtyDog Defense Force
What will be the actual real world performance advantage that PS5 ssd will offer compared to series X. Is it just 1 second loading times faster? NVMEs are 7 times faster than regular ssd but loading times are just one second faster.

With the series X its very easy to explain. The graphics fidelity will be the best with series X having the best resolution, graphical effects and ray tracing. Can someone explain this in real world performance what PS5 ssd will offer.


This video is not a good representation of how things will work next gen outside of loading a game.

The main focus here is data streaming.

I'll give an example:

Lets say a level called "House" was created on game called "Sims" using next-gen hardware. The entire game is in a first-person perspective. The entire level (Note: I said the entire level) was loaded in 1 second on the PS5 and 2 Seconds on the XSX. The total amount of space it took up on Memory was 13.5GB

That should be enough, right?

No.


If the entire level was loaded on to memory in one or two seconds, then that would mean there needs to be lower quality objects and textures to make it fit in that 13.5GB worth of Memory.

If devs want to make the game look better, then they can stream data.

Instead of loading the entire level, they can stream data as soon as you step into a room. That means the data can load so quality that each room or area (kitchen, bathroom, garage etc) can have 5GB worth of data.

This would not be possible on current gen consoles because large sets of data cannot be streamed this fast using a standard HDD.

How does this relate to XSX and PS5 speeds?

The XsX may not be able to stream data that fast, so they would either have to reduce the size of some textures (reduce the load) to keep up with the PS5's streaming speed.

Data needs to be loaded within 0.5 seconds. If PS5 can do it within that time frame and the XsX can't, then you have to reduce the size (some texture quality) to keep up with the PS5's SSD speeds .
 
It's not a lie.

People were posting that Spider-Man GDC video several times just to show people how SSDs will improve next gen graphics.

The Spider-Man demo was only meant to show off the streaming of a specific gameplay slice/section. Conversely there have been people taking the SOD2 load time on XSX as indicative of XSX texture streaming rate which is false, as that was a demo to load the entire game.

The thing is, XboxGAF have been accusing Digital Foundry, NX Gamer etc. of downplaying TF.

If the PS5 was 12.2TF and the XSX was 12.1, people would still be talking about the SSD and how much it would improve games.

They would, but not to the fervent levels some are at current time. It's funny to point out Xbox people accusing DF of downplaying anything, because for a long time it was some of the hardcore Sony fans insisting DF were MS shills and spreading FUD about PS. That all conveniently stopped after they posted their PS5 video.

I wouldn't call it a narrative shift since people are basically learning from what DF and others are teaching them.

Because they are more willing to accept that now that the battleground they were obsessed to the extreme prior (TFs), has been "lost". Again, other people were saying a lot of the same things DF said but these same people were ignored to made to look like fools.

And that is precisely because some of the people now conveniently willing to listen to DF, NX Gamer, Coreleks etc. are doing so partly to continue waging a battle they already started way earlier, just from a different front. There are some taking what these others are saying to heart, but all the same there are some doing it with intent to just continue console warrior bullocks they were already doing well earlier.

CPU performance is just under 3%
GPU performance is around 15%
Memory is going to depend how it's used. Will PS5's setup be different than XsX, whic will allow more ran to be use dfor games? Still remains to be seen.

Percentages don't mean everything. There's a reason I mentioned GPGPU programming earlier. Look at what Sony's own 1st party games were able to do with a paltry 300 - 500 GFLOPs of GPU headroom on base PS4. And a lot of that was done to make up for the crummy Jaguar cores.

Now you are talking about a system this upcoming gen with 1.87 TF - 2 TF of raw GPU headroom over its competitor, on equivalent architecture, and an industry of developers more used to GPGPU task offloading than they were several years prior, with more scalable engines and engine components, and more advanced/sophisticated and lower-overhead programming and scripting techniques tuned to GPGPU tasks.

That is why simply looking at the percentages as being "not that impressive" in light of the SSD delta (on paper) is a bad way of looking at things. And I will remain insisting that GPGPU programming has much more potential for design paradigm shifts in game design than the SSDs, so much evidence and results already point to this. Not me downplaying the SSDs whatsoever, but it is what it is.

So that leaves the SSD, which is about 126% difference

And the context of that delta? This is like saying Jerry ran the mile in 55 minutes and Thomas ran the mile in 25 minutes. That isn't saying much when the passing score is usually 6 minutes xD.

Again, there's a lot to the SSDs we don't know about yet but people should probably temper their expectations some when it comes to how much a major factor they will play in any paradigm shift. And this applies to both consoles.

From what I can tell, XboxGAF is treating the GPU and CPU specs are as something HUGE, but somehow think a 126% increase of raw power will not do much for next gen games?

Already answered this in earlier parts of my comment, no need to go over it again.

People didn't know how fast the SSD was in comparison to the XSX because there was no conformation. When more people started to speak out (NX Gamer, Moore's Law is Dead etc.) that's when people starts to fully realize what the SSDs were capable of.

Both of whom also conveniently saw the next-gen systems in a given way based off of that, conforming to what some of the hardcore Sony fans want to perceive things as. I'm not actually criticizing NX Gamer whatsoever here, because their analysis are generally super-good and I agreed with a lot of things they brought up both in the SSD and GPU vids they've done so far. MLID, however, is a bit more questionable because he made several bad assumptions on design decisions relating to XSX in his video, some of which he contradicted only minutes earlier in the same video!

Still though, that people are more open to learn about the SSDs and speculate what they're capable of from listening to people such as them is great! The problems (well, not even necessarily "problems", per se) come in more when people who aren't considering the full scope of inherent limits to NAND, how flash memory controllers work, aspects of the controllers already mentioned (for example having the idea that the SRAM cache is the same size as a DRAM cache when that is not the case) etc., and therefore overestimate what the SSDs will actually be capable of and what use-cases they will be most suited towards.
 
Last edited:

LED Guy?

Banned
My gripe is this:

Yes, the PS5 SSD is much much faster, but in practice how much useful will it be? The analogy I always like is two cars are racing, one car at 200 mph and the other at 100 mph. It’s a huge difference, but both cars only need to travel 25 feet. We are talking about a blink of an eye, basically.
I think we'll see games getting the most out of the system and you'll be seeing games loading better and having higher resolution textures in the future, like 3 or 4 years into the generation.

And do not worry, you'll see Sony exclusives will be built only on PS5 and you'll see stuff that I don't think it'll be possible on XSX, but that's what the numbers suggest.
 

Elenchus

Banned
I'm curios what the "real" numbers for those SSDs are, a.k.a. random read speeds... Nevertheless I expect that there won't be any forced/prolonged cut-scenes, scripts, elevators, staircases etc. anymore thanks to faster data streaming, many call it a game-changer, a new paradigm in design, whatever drives their boat, but it will a nice change indeed.

Call me crazy but I like the little translations in GOW when Kratos would duck under a rock or squeeze between a gap. Running around the tree on the realm jumps can go but I don’t mind the rest as much as some here.
 

Ascend

Member
This video is not a good representation of how things will work next gen outside of loading a game.

The main focus here is data streaming.

I'll give an example:

Lets say a level called "House" was created on game called "Sims" using next-gen hardware. The entire game is in a first-person perspective. The entire level (Note: I said the entire level) was loaded in 1 second on the PS5 and 2 Seconds on the XSX. The total amount of space it took up on Memory was 13.5GB

That should be enough, right?

No.


If the entire level was loaded on to memory in one or two seconds, then that would mean there needs to be lower quality objects and textures to make it fit in that 13.5GB worth of Memory.

If devs want to make the game look better, then they can stream data.

Instead of loading the entire level, they can stream data as soon as you step into a room. That means the data can load so quality that each room or area (kitchen, bathroom, garage etc) can have 5GB worth of data.

This would not be possible on current gen consoles because large sets of data cannot be streamed this fast using a standard HDD.

How does this relate to XSX and PS5 speeds?

The XsX may not be able to stream data that fast, so they would either have to reduce the size of some textures (reduce the load) to keep up with the PS5's streaming speed.

Data needs to be loaded within 0.5 seconds. If PS5 can do it within that time frame and the XsX can't, then you have to reduce the size (some texture quality) to keep up with the PS5's SSD speeds .
Theoretically and in isolation, yes.

In practice, it doesn't really work that way. Everything that is in the RAM needs to be transferred to the GPU to be displayed. The bandwidth to the GPU will become the limiting factor of how much textures can be displayed, rather than the SSD speed. The normal route is SSD -> RAM -> GPU -> Display (extremely simplified). The reason that RAM is there, is because storage is slow. If SSD was so great to be able to boost texture streaming beyond RAM, RAM would no longer be necessary. An SSD can assist with RAM capacity, because streaming in and out of RAM can happen faster compared to an HDD. So in practice, the PS5 can have an advantage in RAM usage at any given point, but it does not necessarily give superior textures on your screen, because the bandwidth between the RAM and the GPU is still a factor.
There is also the possibility of being able to stream textures directly from the SSD to the GPU, but the SSD throughput is basically negligible compared to the RAM bandwidth.
 

Journey

Banned
The XsX may not be able to stream data that fast, so they would either have to reduce the size of some textures (reduce the load) to keep up with the PS5's streaming speed.

Data needs to be loaded within 0.5 seconds. If PS5 can do it within that time frame and the XsX can't, then you have to reduce the size (some texture quality) to keep up with the PS5's SSD speeds .


I think this is where texture compression comes in. PS5 is using Kraken which is good, but not as efficient as MS' solution. MS is using an entirely cutting edge compression algorithm (BCPack) that unlike Kraken which is general purpose; is specifically designed for game textures.

- Kraken: Reduces the size of a complex non-RDO encoded BC7 format texture (say a normal map) by approx. 20-30%.
- BCPack: Approx. 50+% size reduction. Depends on how far MS pushed the tech. Definitely more effective than just Kraken alone.

Keep in mind that both of these algorithms use "Lossless Compression" so there's not loss in quality.

Next gen will be interesting indeed.
 
Last edited:

ZywyPL

Banned
That's false.

ps5-slides-06-640x360.png


"Contains data for next 1 second of gameplay"

That's how much gameplay will be there between the cut-scenes ;P

[...] but somehow think a 126% increase of raw power will not do much for next gen games?

Here's the thing - that extra GPU and CPU power will undoubtedly have an impact on 100% of games. On a contrary, on how many % of the games that extra SSD bandwidth will change anything but slightly shorter loading times? My personal bet? 0%, aside Sony's exclusives that will most likely do some sick stuff with it. Which unfortunately will probably be just 5-6 titles released between 2023-2028, given how little studios Sony has and how long it takes them to make a game.
 

mckmas8808

Mckmaster uses MasterCard to buy Slave drives
The “SSD secret sauce” talk started when people tried to claim the PS5’s storage bandwidth would single handedly allow for for better graphics and higher FPS.

As far as compression and whatnot, that’s not secret sauce. Correct me if I’m wrong, but the one issue with compressed data is takes more CPU usage to handle, basicslly?

Without compression, both consoles are still extremely fast. PS5 much more so but the XsX SSD will not be a bottleneck of any sort.

Im really excited for an all SSD future on consoles but I think people are building them up to be something they aren’t. They are a big deal, but they only account for so much of what you experience in game.

The secret sauce term is just STUPID. But you are right that with more compression it uses more CPU usage. Unless.......you add more hardware in the box to handle that decompression, then you're fine.
 

mckmas8808

Mckmaster uses MasterCard to buy Slave drives
So Gaming Bolt is claiming PS5 will run most games in 1800p and that XSX’s compression can allow up to 100gb to be streamed in? That’s insane.

I get that PS5 can stream in even more but really, when would you ever need more than 100gb for a single scene?

Sony needs to get price down to $400. Just eat the costs and get it back from services. You cannot allow price parity between these consoles.

I wish you were smarter than this. You don't really believe this do you?
 
I'm curios what the "real" numbers for those SSDs are, a.k.a. random read speeds... Nevertheless I expect that there won't be any forced/prolonged cut-scenes, scripts, elevators, staircases etc. anymore thanks to faster data streaming, many call it a game-changer, a new paradigm in design, whatever drives their boat, but it will a nice change indeed.

This might a case where sometimes less is more. I don't know if gaming as a medium is ready to discuss this kind of stuff yet, but sometimes you can use these things we might perceive as flaws, to accentuate the game design and storytelling.

Look at films for example. Pulp Fiction has quite a few scenes that other films would've cut down dramatically or cut altogether (Butch walking to his apartment, for instance). But those scenes are vital to the film's pacing and storytelling.

Sometimes flaws, or techniques used to mask perceived flaws instead of outright eliminate them, can actually give something personality. I hope developers keep that in mind and don't liberally use every new feature in next-gen simply because it's new. If it serves the game design and storytelling, use it. If it doesn't, might want to think on it.

because after the hdd has loaded the game into ram, it's advantage ends. All this SSD, SSD, SSD shit, have you been living under a rock?

I wouldn't go that far. Both systems are taking an approach (albeit scaled down) similar to AMD's SSG line. Those are workstation cards that feed giant pools of NAND to the GPU directly for streaming of texture data. You can find videos on YT that show the advantages of this type of tech and implementation.

So no, the SSDs will have more uses outside of quickly loading game data. However, some people aren't understanding to what degree that texture streaming will work. The idea that texture streaming from NAND will be comparable to the GPU working on data from RAM is ridiculous; not only because GDDR6 is much faster and provides much more bandwidth, but because the level of granularity is much smaller (and the smaller that granularity, the better that data is for modification, alteration, etc).

Since NAND data has magnitudes larger granularity, unless there's a step in the pipeline to break that data up into smaller chunks for the GPU to work on (this is part of what I suspect the SRAM cache in PS5's memory controller is for), that significantly limits the application of data coming off the NAND. Not to mention, PCIe has its own way of packaging data and sending it along the lanes that's much different from RAM memory controllers.

Just 'little" things like that, I feel are worth keeping in mind to manage expectations of what areas the SSDs will be particularly beneficial with in actual practice. But yeah, they will bring more to the table than just quicker loading times.
 
Last edited:

SonGoku

Member
This article is full of misinformation, like it was written by a xbox fan completely oblivious to the tech inside PS5
Percentages don't mean everything. There's a reason I mentioned GPGPU programming earlier. Look at what Sony's own 1st party games were able to do with a paltry 300 - 500 GFLOPs of GPU headroom on base PS4. And a lot of that was done to make up for the crummy Jaguar cores.
Neither console can go above its theoretical peak (10.27TF & 12.1TF)
Optimizing for asynchronous compute can only help XSX get closer in translating its compute advantage (17-21%) to in game performance linearly

PS4-XB1 gap was twice the gap between PS5-XSX. Its much closer
And I will remain insisting that GPGPU programming has much more potential for design paradigm shifts in game design than the SSDs,
A fully realized 17-21% compute performance advantage will translate to 17-21% higher resolution best case scenario at exact same settings. Especially in the region of 4K this delta is not significant.
Unless you meant GPGPU programming for next gen in general.
 
Last edited:
What will be the actual real world performance advantage that PS5 ssd will offer compared to series X. Is it just 1 second loading times faster? NVMEs are 7 times faster than regular ssd but loading times are just one second faster.

With the series X its very easy to explain. The graphics fidelity will be the best with series X having the best resolution, graphical effects and ray tracing. Can someone explain this in real world performance what PS5 ssd will offer.

About the fastest speed I've seen on pc ssds is 5x hdd. Why that is? Could be the bottlenecks Cerny removed on ps5 or something else. The ps5 is 100x hdd speed, or 20x faster than the fastest pc ssd speed seen.
but it does not necessarily give superior textures on your screen, because the bandwidth between the RAM and the GPU is still a factor.
There is also the possibility of being able to stream textures directly from the SSD to the GPU, but the SSD throughput is basically negligible compared to the RAM bandwidth.

The problem is you're assuming textures are already using the maximum capacity of the ram's bandwidth. That isn't the case, even on ps4 there are some textures that are ultra sharp, about as sharp as the eye can discern, and moments later you see textures substantially lower detail in another area. Sometimes you get even blurry textures.

The ps5 has significantly higher bandwidth than the pro, and the pro already can handle extremely sharp textures.The ps4 system was shown to be able to display ultra sharp textures, but clearly the dev.s prioritized certain objects and areas over others. What the exnaughty dog developer mentioned was being able to have very high highest level of detail, so that as you approached objects they'd have the highest detail possible. Which can be alot. As you'd only be near a few objects in the scene, those objects which would occupy most of the screen can benefit from additional detail that the bandwidth can allow(when you got further away they'd be replaced by lower level of detail, and need substantially less bandwidth).

We've all seen how certain models and certain textures in games look substantially better than most everything else in the game. Why? Because developers focused on those models and textures, they prioritized them, and they had to downscale other objects and textures to not affect loading or streaming.
Keep in mind that both of these algorithms use "Lossless Compression" so there's not loss in quality.
Are you sure BCPACK is lossless?
 
Last edited:
This article is full of misinformation, like it was written by a xbox fan completely oblivious to the tech inside PS5

Neither console can go above its theoretical peak (10.27TF & 12.1TF)
Optimizing for asynchronous compute can only help XSX get closer in translating its compute advantage (17-21%) to in game performance linearly

PS4-XB1 gap was twice the gap between PS5-XSX. Its much closer

I've never argued that PS5 and XSX aren't closer in capability compared to PS4 and XBO. The point was more to show what PS4 was able to accomplish with its GPU headroom over XBO in terms of pure theoretical compute amount, which was roughly 300 - 500 GFLOPs of GCN architecture.

If a game on PS5 and XSX target identical graphics fidelity, XSX has additional headroom for offloading of non-graphics tasks on the GPU that the PS5 would not have unless it aimed for lower comparative graphical fidelity. And that's thanks to the XSX having 16 additional CUs on the GPU. This is just in reference to tasks being done in parallel, no clock differential factoring.

Which, while useful, kind of defeats the purpose of looking at the two comparatively in that regard. It's an area XSX has an obvious advantage in, particularly factoring into account increased efficiency/optimization of techniques aimed at asynchronous compute, as well as more developer familiarity with those techniques (this was the first gen to really push it, a lot due out of necessity of the underperformance of the Jaguar cores). The GPU advantage may only be 17% - 21% between the systems this time around, but asynchronous compute techniques and engine compartmental scalability have improved significantly over the past seven years as well.

Not even just that, but the ability to saturate the CUs with more tasks efficiently in parallel is also possible due to RDNA2 improvements and developer growth with APIs targeting that type of work. Which is why, ultimately, percentages of paper specs don't tell the whole story. FWIW this is usually extended as givens when speaking about the SSDs, even if I'm one of the people asking questions on other parameters of the SSDs that haven't been divulged (asking for good reason IMO). Some people are simply looking at the percentage delta but don't take enough into context of what those percentages are in regards to and what role in terms of overall system performance those things actually play.

General rule is, the CPU and GPU are always the brain and heart of a console. The RAM can be considered the stomach, maybe? And the SSDs are more like the intestines. Their role is still vital, but not on the same level as the other three components. That won't change with next-gen, either, though at very least we will finally get to see storage used much more efficiently and to its potential, and that counts for a lot.

About the fastest speed I've seen on pc ssds is 5x hdd. Why that is? Could be the bottlenecks Cerny removed on ps5 or something else. The ps5 is 100x hdd speed, or 20x faster than the fastest pc ssd speed seen.

Actually it's 500 MB faster than the fastest PC SSD currently on market, the Sabrent Rocket Gen 4.

BCPack may or may not be lossless, but that doesn't necessarily matter since they have a lossless compression format alongside it. BCPack is aimed specifically for texture compression and it seems it will have features on XSX not available on PC more optimized/specialized for the console.
 
Last edited:
Actually it's 500 MB faster than the fastest PC SSD currently on market, the Sabrent Rocket Gen 4.

BCPack may or may not be lossless, but that doesn't necessarily matter since they have a lossless compression format alongside it. BCPack is aimed specifically for texture compression and it seems it will have features on XSX not available on PC more optimized/specialized for the console.
Yes but the sabrent is nearly 100x faster than an hdd, but you don't see that in practice. As far as I can see it is about as fast as a sata ssd when loading games, and about 2-5x faster than an hdd depending on game.

The ps5 ssd is 100x faster than an hdd, not 2-5x faster. Had Cerny not removed the bottlenecks it would never be 100x despite the specs.

Why are you seeing loading up to 30+ seconds depending on the game on the sabrent? Could be the bottlenecks, the bottlenecks the ps5 does not have. It is conceivable the ps5 could load the same games in 1 to 2 seconds.
 
Last edited:

Ascend

Member
The problem is you're assuming textures are already using the maximum capacity of the ram's bandwidth.
Go talk to developers, and you'll know that the main thing that has evolved the slowest over time is bandwidth. CPUs have grown exponentially in performance, and so have GPUs. But buses are still the slowest things around relatively speaking. There's a reason AMD came up with HBM, and also that we're constantly increasing both cache sizes and amount of cache levels in both CPUs and GPUs. Those are all there to mitigate the obvious issue of bandwidth restraint between the main components of computing systems.
So it is not an assumption in that sense. And I'm definitely not assuming that textures specifically are using the max capacity of the RAM's bandwidth. Everything needs to be transferred through that same bandwidth after all, so they are simply part of the whole package, although textures are obviously one of the larger ones in terms of raw size. But why would one use anything else other than max bandwidth? Using anything less means slowing everything down unnecessarily.

That isn't the case, even on ps4 there are some textures that are ultra sharp, about as sharp as the eye can discern, and moments later you see textures substantially lower detail in another area. Sometimes you get even blurry textures.
And why would that be...?

The ps5 has significantly higher bandwidth than the pro, and the pro already can handle extremely sharp textures.The ps4 system was shown to be able to display ultra sharp textures, but clearly the dev.s prioritized certain objects and areas over others. What the exnaughty dog developer mentioned was being able to have very high level of detail, so that as you approached objects they'd have the highest detail possible. Which can be alot. As you'd only be near a few objects in the scene, those objects which would occupy most of the screen can benefit from additional detail that the bandwidth can allow.
That sounds more like variable rate shading, which is removing the need to load sharp textures for 'far away' objects that don't have many pixels anyway... This was mostly done manually in the past, but now the consoles support that in-hardware without much programming needed from developers. And why would one do such a thing? To decrease the work load on the GPU, but more importantly, to lessen the streaming of textures that won't make a difference visually on screen, but still cost bandwidth.

We've all seen how certain models and certain textures in games look substantially better than most everything else in the game. Why? Because developers focused on those models and textures, they prioritized them, and they had to downscale other objects and textures to not affect loading or streaming.
You just said it yourself in bold... Loading or streaming, which, is determined by two things. Speed of the storage and bandwidth. And I already explained that storage transfer speed is MUCH slower than RAM speed... The advantage of SSD over HDD is that more can be sent into the RAM in a much shorter time, meaning, your RAM allocation can be less, reducing the amount of RAM you need at any given moment. This can free up RAM for other things, which ultimately can allow for more things in the game. So if with an HDD you need 10GB of RAM, with an SSD you maybe need 7GB of RAM (these numbers are hypothetical). So theoretically, you could load up an additional 3GB of textures and have the same performance of when you were running an HDD. But if you bandwidth could only handle an additional 1GB, those additional 2GB would not only be a waste, it would still tank the performance.
 
Yes but the sabrent is nearly 100x faster than an hdd, but you don't see that in practice. As far as I can see it is about as fast as a sata ssd when loading games, and about 2-5x faster than an hdd depending on game.

The ps5 ssd is 100x faster than an hdd, not 2-5x faster. Had Cerny not removed the bottlenecks it would never be 100x despite the specs.

Why are you seeing loading up to 30+ seconds depending on the game on the sabrent? Could be the bottlenecks, the bottlenecks the ps5 does not have. It is conceivable the ps5 could load the same games in 1 to 2 seconds.

You're going off a lot of blind faith on what Cerny said. And let's be fair here, Cerny embellished a few things in his presentation, since he was also serving as PR for PS5 during that speech. How do we know for sure Sony's removed these bottlenecks? Do we just take their word on it? Why not wait until we see the systems in real-time before stating these kind of claims absolutely?

What do you mean you "don't see that in practice"? when it comes to the Sabrent drive? Magazines and other places test these cards all the time and if they couldn't get the performance the manufacturer claims, they'd have a PR disaster and possibility even be liable for false advertising. Again, you take Sony at their word blindly but cast doubt on Sabrent even though I doubt you have used their drive, but other reviewers who have are making things up? We haven't even seen Sony's SSD in action; MS's either for that matter but at least they are stating guaranteed performance. We'll see how that holds up in practice but it's at least something.

For as much as people bring up the bottlenecks supposedly removed, they aren't being very clear on what the bottlenecks were in the first place. It's no secret that games on PC never targeted SSDs as an expected feature to begin with; if that is one of the "bottlenecks" then it's solved simply with more PC games natively targeting SSDs, like Star Citizen. Is lack of cache one of the bottlenecks? Because high-tier SSDs have usually had both DRAM caches (up to 128 MB - 256 MB, some even 512 MB and I wouldn't be surprised if some had more. Those are all capacities an SRAM cache cannot realistically match BTW), but also a small portion of faster SLC NAND, all structured as a hierarchy?

Is it the file I/O system? How do we know what that will look like on PC going forward? How do we know MS haven't done something similar? It can be gleamed from their own talk that at least some measures have been taken in this regard, but those seem to routinely go unnoticed (maybe because it doesn't easily confirm to the black & white narrative? Which I'm using "narrative" in the storytelling term, btw).

Too many questions, too little real data, zero real-life results to gauge on to just blindly take Cerny's words for what they are. Yes, we can give the benefit of the doubt but a little informed skepticism is never a bad thing. People seem loving to do that all the time when it comes to MS, but it's somehow implied taboo if you do the same with Sony and that is more than a little weird.

It's an ASIC, and it's not in XBSX. You can move along...

Is that you, Matt?

The ASIC itself doesn't need to be present for the technology therein to be used in some capacity in another device. We don't know everything on the hardware anyway, let's leave some room here to speculate 👍
 
Last edited:
Too many questions, too little real data, zero real-life results to gauge on to just blindly take Cerny's words for what they are. Yes, we can give the benefit of the doubt but a little informed skepticism is never a bad thing. People seem loving to do that all the time when it comes to MS, but it's somehow implied taboo if you do the same with Sony and that is more than a little weird.

I wish this was stickied to the top of the forum
 
But why would one use anything else other than max bandwidth? Using anything less means slowing everything down unnecessarily.
Because hdd is 100MB/s do you prioritize the highest lod of an object that the player will interact with, or some random backdrop? You can't have both with similar highest LOD without affecting loading significantly, or exceding the capacity of the streaming bandwidth from hdd.

And why would that be...?

That sounds more like variable rate shading, which is removing the need to load sharp textures for 'far away' objects that don't have many pixels anyway... This was mostly done manually in the past, but now the consoles support that in-hardware without much programming needed from developers. And why would one do such a thing? To decrease the work load on the GPU, but more important
Those textures still need to come from the hdd in the current gen. The geometry also. There is various levels of detail, and now variable rate shading. But whenever you approach an object the highest lod has to be in ram. Theoretically a large room could have 20+GB of highest LOD for when you approach objects, but that would have to be streamed in and out as you approached objects.
You just said it yourself in bold... Loading or streaming, which, is determined by two things. Speed of the storage and bandwidth. And I already explained that storage transfer speed is MUCH slower than RAM speed... The advantage of SSD over HDD is that more can be sent into the RAM in a much shorter time, meaning, your RAM allocation can be less, reducing the amount of RAM you need at any given moment. This can free up RAM for other things, which ultimately can allow for more things in the game. So if with an HDD you need 10GB of RAM, with an SSD you maybe need 7GB of RAM (these numbers are hypothetical). So theoretically, you could load up an additional 3GB of textures and have the same performance of when you were running an HDD. But if you bandwidth could only handle an additional 1GB, those additional 2GB would not only be a waste, it would still tank the performance.
Take a city like spiderman's. Theoretically it could use 100GB of highest LOD objects for the whole city. As you moved about you could stream 8-10~+GB of data for nearby objects. In the ps4 they already said they used blurrier textures in spiderman because the streaming speed was insufficient for higher lod. In ps5 all objects can be highest lod if the developer has the development man power to do so.

That's the thing variable level of detail and variable rate shading, means only objects nearby and occupying large portions of the screen can benefit from most of the bandwitdth to allow for extremely high LOD.

Look at this demo for example.


Clearly the ps4 can handle assets of that level of detail, as it is doing so. But if all scenes were with that level of detail each room would likely need 1+minute load times. On ps5 the streaming bandwidth is no longer an issue, and LOD can be as high as the new hardware can support without issue.
 

psorcerer

Banned
Go talk to developers, and you'll know that the main thing that has evolved the slowest over time is bandwidth.

Latency. Bandwidth is pretty ok. Doesn't grow as fast as ALUs, but it grows.
Latency only increases. It never gets faster, only slower.

But if you bandwidth could only handle an additional 1GB, those additional 2GB would not only be a waste, it would still tank the performance.

Better textures do not tank the bandwidth the way you describe.
Wast majority of bandwidth is spent on writing the various render results (render targets, shadow maps, buffers, etc.)
Reading textures is not a big bandwidth hog.
 

SonGoku

Member
I've never argued that PS5 and XSX aren't closer in capability compared to PS4 and XBO. The point was more to show what PS4 was able to accomplish with its GPU headroom over XBO in terms of pure theoretical compute amount, which was roughly 300 - 500 GFLOPs of GCN architecture.
But the gap was bigger... XSX doesn't have the same headroom its advantage will be much less apparent than PS4/XB1 and that's before factoring diminishing returns from resolutions in the region of 4K, you can't possibly expect the same difference in output. Dismissing percentages to focus on flops differences between generations (500GF & 2TF) is meaningless without taking into account proportions
If a game on PS5 and XSX target identical graphics fidelity, XSX has additional headroom for offloading of non-graphics tasks on the GPU that the PS5 would not have unless it aimed for lower comparative graphical fidelity. And that's thanks to the XSX having 16 additional CUs on the GPU.
When i said parity i meant everything from visuals to physics, particle effects etc. All it would take to produce the same results (visual and compute oriented) is to run at 17-21% lower resolution
Which, while useful, kind of defeats the purpose of looking at the two comparatively in that regard. It's an area XSX has an obvious advantage in, particularly factoring into account increased efficiency/optimization of techniques aimed at asynchronous compute, as well as more developer familiarity with those techniques (this was the first gen to really push it, a lot due out of necessity of the underperformance of the Jaguar cores). The GPU advantage may only be 17% - 21% between the systems this time around, but asynchronous compute techniques and engine compartmental scalability have improved significantly over the past seven years as well.
Asynchronous compute techniques will also benefit PS5 though... The best case scenario for XSX with asynchronous compute is it'll reach the same level of GPU utilization as PS5 which is again a 21% gap at best. Running at 21% lower resolution would ensure it can reach settings parity
 
Last edited:
" then it's solved simply with more PC games natively targeting SSDs, like Star Citizen.
star citizen doesn't load much faster than from an hdd, think it was 2x faster. Certainly nowhere near 100x faster on an nvme.



On an ssd it seems to take over 1 minute to load vs 2 minute on hdd. On ps5 if we believe Cerny that might easily be 1.2~ seconds.

You say it's blind faith on Cerny. But he clearly mentioned several of the bottlenecks, the custom i/o h/w, probably including some asics in the i/o of the apu. Not only that but he said developers can see it in practice. Already the spiderman demo gave a 10x speed up. Imagine 10x speed up on star citizen, it would go from 140 seconds load time on hdd to 14 seconds. But the spiderman demo loaded in under 1 second, there is likely some minimal scene setup time that will make loading at best at least take a significant fraction of a second.
 
Last edited:
This is a lie. That is the point a good portion of posters have been trying to argue, generally by misinterpreting how SSDs and NAND actually works. It hasn't been so much some people downplaying SSDS, so much as anyone attempting to be a realist with regard to the SSDs is automatically viewed by some others as downplaying the SSDs.

Let's face it; after the Road to PS5 presentation (and I don't like to do this but a spade's a spade) a big flock of Sony fans on the forums who were obsessed to high hell over Teraflops (even when multiple posters, myself included, were trying to tell them Teraflops didn't mean everything), silently conceded that front. They began downplaying the Teraflop difference between the two systems (not in terms of percentages per-se, but in what the extra TF advantage on XSX can actually be utilized for) while hyping up the SSD and audio since Sony focused on those in particular with their presentation and gave specs that, on paper, seemed more impressive than MS's in those areas. It allowed those people to shift the narrative to the SSD, audio, I/O etc. while similarly creating a fake narrative of XSX "brute forcing" a solution and PS5 was the system pushing elegant optimizations, conveniently cutting out any focus on MS's deliberate optimizations and customizations with the XSX to further push this fake narrative.

All the while, many of these same people continue to over-inflate the SSDs in terms of being a game-changing technology or paradigm shift, and completely downplay the GPGPU performance edge XSX has over PS5 (or pretend it doesn't exist at all and that the extra GPU throughput of XSX will only go to resolution, ignoring the ML texture upscaling in the GPU built to free up heavy expenditure of GPU resources on processing raw pixels through to the display, freeing up processing power for other tasks). When you try telling them that NAND has limits inherent to the technology that will prevent granularity of asset data for streaming in a way similar to volatile RAM, somehow that gets lumped into "downplaying the PS5 SSD", even though this affects both systems. Same if you bring up questions regarding the random write speeds, latency figures, page and block sizes, etc.

We are seemingly allowed to speculate on Sony using tech from other departments of their company as R&D foundations for potential PS5 features, but doing the same with MS regarding XSX is considered being a fanboy, wishful thinking, or foolish...even though they have already admitted to members of the Surface team working on the Xbox team. All the same, some strongly pro-Sony people who obsess over customizations on PS5 do not provide any leeway to entertain similar customizations conceptually being present on XSX, but expect strong pro-Microsoft people to bend the knee and do so when it comes to XSX features potentially being present on PS5. And all of this leads to disingenuous, lopsided, biased takes and discussions in next-gen speculation because there are a group of people who put out a false image of wanting the best for both systems but secretly only want their preferred platform to "win", even if that means generating fake narratives.

Yes there are some Xbox people who do this but from what I've noticed it is not to the same degree as the Sony fans engaging in similar tactics on the forum (as just one example). Now that might be going a bit beyond your point here but it needs to be stressed that claiming "When people tried to explain how SSDs will work, people just started saying, "It's not going to close the power gap in consoles" when that wasn't the point people on here were trying to make." is in fact demostrably false when keeping in mind the long-term discussion that's been prevalent for months by now on these systems.

There absolutely have been people trying to imply this very thing, maybe not directly and often layered in subtext, but it's been an idea fostered for a good bit by now. Again, predicated on things like "secret sauce" like the way this article words it, which is irresponsible considering so many important aspects of the systems are not even divulged yet. But that's all I want to say on that and thought what you mentioned was a good time to segway into it. Hopefully people get what I'm saying here.
The fact someone instantly argued against this, SMH this forum. Nonsensically shouting “SSD!” and “Bottlenecks!” ad nauseam to defend their chosen plastic box.
 

SonGoku

Member
What do you mean you "don't see that in practice"? w
I assume he means games designed around HDD wont take full advantage of its read capabilities during loading and further more it won't change the design decisions made with HDD in mind
Im sure you agree its a pointless comparison which is not representative of the differences the SSDs inside consoles will make for games designed around that spec
For as much as people bring up the bottlenecks supposedly removed, they aren't being very clear on what the bottlenecks were in the first place. It's no secret that games on PC never targeted SSDs as an expected feature to begin with; if that is one of the "bottlenecks" then it's solved simply with more PC games natively targeting SSDs, like Star Citizen. Is lack of cache one of the bottlenecks?
PS5 I/O has dedicated hardware to stream 8-9GB worth of compressed data typically, to do the same on PC without bespoke hardware they'll need to dedicate the equivalent of 11 ZEN2 cores
Of course PC doesn't need as high streaming speeds from SSD since they can make up for it with copious amounts of DDR4 RAM that act as a buffer
 
Last edited:
A relevant quote from another thread on ps5 ssd benefits

False?

In Theory: How SSD Could Radically Change Next-Gen Games Beyond Faster Loading


The final part of the video talked about this fidelity and scope and how it would only be possible with the SSD.



But we can't store all of the super detailed high res versions for all objs in memory at once?



Doesn't seem false to him, neither does it seem false to DF. It may not be life like, but it clearly appears better based on what they were able to do on a regular slower storage device.

Particularly note this bit
Even with current gen hardware we can render most single objects at lifelike detail... But we can't store all of the super detailed high res versions for all objs in memory.-Andrew Maximov. Technical Art Director, Naughty Dog. Maximov is the technical art director at Naughty Dog[former?], the lauded creators of blockbuster titles like 2016's Uncharted 4.
 

pawel86ck

Banned
What will be the actual real world performance advantage that PS5 ssd will offer compared to series X. Is it just 1 second loading times faster? NVMEs are 7 times faster than regular ssd but loading times are just one second faster.

With the series X its very easy to explain. The graphics fidelity will be the best with series X having the best resolution, graphical effects and ray tracing. Can someone explain this in real world performance what PS5 ssd will offer.

SDDs on PC are bottlenecked because of decompression speed so that's why there's very little difference between slow and fast SDD.
 

SonGoku

Member
SDDs on PC are bottlenecked because of decompression speed so that's why there's very little difference between slow and fast SDD.
and if the game wasn't designed to load from storage past a certain speed an even faster SSD won't make a difference
 
Last edited:

DForce

NaughtyDog Defense Force
That's how much gameplay will be there between the cut-scenes ;P



Here's the thing - that extra GPU and CPU power will undoubtedly have an impact on 100% of games. On a contrary, on how many % of the games that extra SSD bandwidth will change anything but slightly shorter loading times? My personal bet? 0%, aside Sony's exclusives that will most likely do some sick stuff with it. Which unfortunately will probably be just 5-6 titles released between 2023-2028, given how little studios Sony has and how long it takes them to make a game.

More downplaying.

2.8% and 18% GPU advantage, "Well undoubtedly have an impact on 100% of games"

Better I\O and 129% more raw SSD Speeds, "It won't change anything with slightly better load times."

You're simply downplaying it. You guys don't want to see PS5 give any advantages. All that RAW power will result in clear advantages, especially when more games start to stream in more assets.


More factions of a second matter. If this also allows PS to have more memory for games (as mentioned by NX Gamer), we could see even more advantages.
 

Ascend

Member
Latency. Bandwidth is pretty ok. Doesn't grow as fast as ALUs, but it grows.
Latency only increases. It never gets faster, only slower.
Depends... For CPU, lower latency is preferred, while for GPUs (and thus gaming consoles), bandwidth is preferred. Without splitting the memory pool, you'd have to choose one over the other, and consoles generally choose bandwidth over latency. This is why the consoles use GDDR6 instead of DDR4. There are indeed workarounds for bandwidth though, like compression. If it wasn't for that, we'd have reached the limit a long time ago. Latency has no work-arounds, so, I understand where you're coming from.

Better textures do not tank the bandwidth the way you describe.
Wast majority of bandwidth is spent on writing the various render results (render targets, shadow maps, buffers, etc.)
Reading textures is not a big bandwidth hog.
Yeah... I don't think I was specific on whether it was reading or writing. Obviously it's the intermediate render results that require being temporarily stored that are going to consume a lot of bandwidth. The rendering itself is obviously not a bandwidth hog. I was trying to explain things simply without making things too complicated.
 

DForce

NaughtyDog Defense Force
The Spider-Man demo was only meant to show off the streaming of a specific gameplay slice/section. Conversely there have been people taking the SOD2 load time on XSX as indicative of XSX texture streaming rate which is false, as that was a demo to load the entire game.

You're doing just like the other guy I had a conversation with. We were NOT talking about the Spider-Man demo Sony showed off, we were talking about the streaming section in the GDC talk video.

Mere seconds mattered and data needed to be streamed within that time frame. Data that was late had to be delayed, which means the speed in streaming played a factor.

They would, but not to the fervent levels some are at current time. It's funny to point out Xbox people accusing DF of downplaying anything, because for a long time it was some of the hardcore Sony fans insisting DF were MS shills and spreading FUD about PS. That all conveniently stopped after they posted their PS5 video.

I'm not accusing anything. Right now you're just making things up. Here's proof.


Wow, look at all the people saying that he's downplaying TFs.

Because they are more willing to accept that now that the battleground they were obsessed to the extreme prior (TFs), has been "lost". Again, other people were saying a lot of the same things DF said but these same people were ignored to made to look like fools.

They wanted to have more TF because of bragging right. People would have celebrated if they had a 0.1 advantage. STILL it wouldn't mean much since they're so close.

And that is precisely because some of the people now conveniently willing to listen to DF, NX Gamer, Coreleks etc. are doing so partly to continue waging a battle they already started way earlier, just from a different front. There are some taking what these others are saying to heart, but all the same there are some doing it with intent to just continue console warrior bullocks they were already doing well earlier.

People have been listening to them for awhile since they can easily breakdown why the SSDs are important.

Don't remember people discrediting devs and calling them all Sony devs when they were praising the PS5 console? Yes, they were all accused of downplaying TF, too.


Percentages don't mean everything. There's a reason I mentioned GPGPU programming earlier. Look at what Sony's own 1st party games were able to do with a paltry 300 - 500 GFLOPs of GPU headroom on base PS4. And a lot of that was done to make up for the crummy Jaguar cores.

Now you are talking about a system this upcoming gen with 1.87 TF - 2 TF of raw GPU headroom over its competitor, on equivalent architecture, and an industry of developers more used to GPGPU task offloading than they were several years prior, with more scalable engines and engine components, and more advanced/sophisticated and lower-overhead programming and scripting techniques tuned to GPGPU tasks.
Headroom? As the generation goes on, that headroom will be a not factor because the GPU is going to be pushed to render pixels. There will be headroom early with both gen consoles as devs might see more than enough GPU power to hit their 4K target.

That is why simply looking at the percentages as being "not that impressive" in light of the SSD delta (on paper) is a bad way of looking at things. And I will remain insisting that GPGPU programming has much more potential for design paradigm shifts in game design than the SSDs, so much evidence and results already point to this. Not me downplaying the SSDs whatsoever, but it is what it is.



And the context of that delta? This is like saying Jerry ran the mile in 55 minutes and Thomas ran the mile in 25 minutes. That isn't saying much when the passing score is usually 6 minutes xD.

Again, there's a lot to the SSDs we don't know about yet but people should probably temper their expectations some when it comes to how much a major factor they will play in any paradigm shift. And this applies to both consoles.



Already answered this in earlier parts of my comment, no need to go over it again.



Both of whom also conveniently saw the next-gen systems in a given way based off of that, conforming to what some of the hardcore Sony fans want to perceive things as. I'm not actually criticizing NX Gamer whatsoever here, because their analysis are generally super-good and I agreed with a lot of things they brought up both in the SSD and GPU vids they've done so far. MLID, however, is a bit more questionable because he made several bad assumptions on design decisions relating to XSX in his video, some of which he contradicted only minutes earlier in the same video!

Still though, that people are more open to learn about the SSDs and speculate what they're capable of from listening to people such as them is great! The problems (well, not even necessarily "problems", per se) come in more when people who aren't considering the full scope of inherent limits to NAND, how flash memory controllers work, aspects of the controllers already mentioned (for example having the idea that the SRAM cache is the same size as a DRAM cache when that is not the case) etc., and therefore overestimate what the SSDs will actually be capable of and what use-cases they will be most suited towards.

If the things you were saying are true, then Sony would've never put much importance on their SSD. I mean, based on what you guys are saying, Sony should've talked to you guys before designing their console.

I said this multiple times that their 5.5GB\s target was to stream data. Not that was their TARGET. That means whatever they wanted to accomplish could have not be done with slower speeds.

Again, you're looking at RAW GPU speeds on paper, yet ignore what's on the PS5.
 

DForce

NaughtyDog Defense Force
I think this is where texture compression comes in. PS5 is using Kraken which is good, but not as efficient as MS' solution. MS is using an entirely cutting edge compression algorithm (BCPack) that unlike Kraken which is general purpose; is specifically designed for game textures.



Keep in mind that both of these algorithms use "Lossless Compression" so there's not loss in quality.

Next gen will be interesting indeed.
And it's not going to make up for RAW speeds which is the point.

People are trying to say this will close the gap.
 

DForce

NaughtyDog Defense Force
Theoretically and in isolation, yes.

In practice, it doesn't really work that way. Everything that is in the RAM needs to be transferred to the GPU to be displayed. The bandwidth to the GPU will become the limiting factor of how much textures can be displayed, rather than the SSD speed. The normal route is SSD -> RAM -> GPU -> Display (extremely simplified). The reason that RAM is there, is because storage is slow. If SSD was so great to be able to boost texture streaming beyond RAM, RAM would no longer be necessary. An SSD can assist with RAM capacity, because streaming in and out of RAM can happen faster compared to an HDD. So in practice, the PS5 can have an advantage in RAM usage at any given point, but it does not necessarily give superior textures on your screen, because the bandwidth between the RAM and the GPU is still a factor.
There is also the possibility of being able to stream textures directly from the SSD to the GPU, but the SSD throughput is basically negligible compared to the RAM bandwidth.

Sure, there are limiting factors, but it also depends on how the data is streamed. Cerny's idea is to stream data as the player is turning. There's only one console who can stream data that fast and there will be limiting factors based on pure speeds. This is also not factoring in their I/O setup which they went in to great lengths to stop any soft of bottlenecks.


But what if they're able to use more memory as NX Gamer suggested? Xbox using a split memory could become a factor as the generation goes on.
 
and if the game wasn't designed to load from storage past a certain speed an even faster SSD won't make a difference
star citizen is designed to run even on titan rtx and 32GB of ram. To me it seems like it should also be designed to benefit from faster ssds. But even it has 1 minute load times on ssd. The bottlenecks Cerny mentioned might be quite real.
 
How about the following quote
However, PS5 devs can encode their textures in a special way (with rate distortion optimization or "RDO"), and this combined with Kraken should make up for the lack of a BCPack equivalent in hardware.-Richard Geldreich?
So ps5 textures would have comparable compression on textures(would quality be affected?), but kraken would also compress all other nontexture data that was needed.
 

ZehDon

Gold Member
For what it's worth, Microsoft's DirectStorage seems like the real-deal: a re-written API designed to make IO interactions faster and more reliable. This is intended to migrate to the Windows platform. So... I can't vouch for how much PR fluff is around the rest of Microsoft's claims, but this at least could be interesting - they wouldn't put it in Windows if it didn't work... well, maybe :messenger_tears_of_joy:
 
Last edited:

Shmunter

Member
The ssd is not so secret, it is a tangible increase in data delivery, of which speed and capability is documented by both holders including compression estimates. Processing and ram speed have exponentially increased through the gens. Ability to feed these monsters has remained mostly stagnant until now. Loading aside, the benefits will be evident in on screen levels of detail.

The secret remains mostly on the Sony side with bespoke processing like the tempest chip and geometry culling processors. These are secret as we are unable to predict real world benefit, if any.
 
Top Bottom